CENOTHERA AND THE MUTATION THEORY 313 
its derivatives rather than to the other species. 
An explanation was suggested in 1914 by Renner, 
who found that about half of the seeds of Lamarck- 
iana fail to germinate. He supposed that the 
species was a permanent heterozygote for the leta- 
velutina pair of characters, which he treated as 
allelomorphs. The homozygous types were both 
assumed to be inviable,.and to be represented by 
the seeds that fail to germinate. This view was at 
first opposed by deVries, but has since been adopted 
by him, and has been substantiated by so much 
evidence that it can now scarcely be doubted. 
By 1917, then, the situation had been analyzed 
far enough to indicate that CEnothera Lamarckiana 
is a permanent heterozygote in at least two respects, 
and that its peculiar behavior in crosses is at least 
in part due to this heterozygosis. But in two 
respects the results were still unsatisfactory. No 
analogous case was known in any other organism, 
so that the hypotheses seemed curiously artificial 
and improbable; and there was no obvious relation 
between the heterozygous nature of O. Lamarckiana 
and the frequency with which it produces new types. 
Both these objections were eliminated by the 
work of Muller on certain peculiar races of Dro- 
sophila. Muller examined the beaded race, which 
had long been a puzzle to those studying Dro- 
sophila. The beaded race, when first studied, did 
not breed true, but constantly produced normal 
flies. After a long period of selection, however, it 
began to breed almost true. In crosses the character 
